A new laser-based loop detection method for laneway environment 3D mapping

This paper presents a 3D SLAM loop detection method for the mapping of laneway environment within lidar scans. Attuned to the fact that laneway settings such as underground mines, long corridors, and long tunnels, are often low-featured narrow tunnel-like environment without GPS availability and poor light condition. The proposed method solves the difficulty of feature extraction in laneway environment by encoding keyframe scans. Furthermore, the similarity score of encoded scans for matching is calculated in an efficient way, which enables the real-time loop detection mechanism while reverse revisiting is encountered in laneway environment. The proposed method is evaluated on datasets gathered from a long corridor environment.